Re: Duplicating from copy of a standby with standby controlfile

From: Ryan January <rjjanuary_at_multiservice.com>
Date: Mon, 10 Feb 2014 17:03:17 -0600
Message-ID: <52F95AB5.10206_at_multiservice.com>

Hi Mark,
Just for clarification, which control file is getting restored on your clone database? Are you backing up the standby control file from the cloned snapshot and restoring it back in place or is the restore a copy of the active control file from the primary database? We're currently doing clones of our production database and I'm looking to migrate these to the physical standby to better utilize it's hardware resources. I would be interested in taking a look at your scripts if you wouldn't mind sending them my way.

Thanks,
Ryan

On 02/10/2014 04:42 PM, Don Seiler wrote:
> Thank Mark. I think Step 6 might be the missing component here for me.
> I'll fuss around with it some more.
>
> We're moving to ZFS and DirectNFS and I know there are a couple of
> snapshot/clone features there that we'll be looking to utilize once
> we're fully migrated.
>
> Thanks,
> Don.
>
>
> On Mon, Feb 10, 2014 at 4:29 PM, Mark Burgess
> <mark_at_burgess-consulting.com.au
> <mailto:mark_at_burgess-consulting.com.au>> wrote:
>
> We do this with a combination of ZFS snapshots and clones from the
> standby database - no backupsets as such.
>
> The method we have implemented to have the correct controlfile in
> place is:
>
> 1. Stop media recovery on the standby database.
> 2. Take the ZFS snapshot and clone of the relevant file systems.
> 3. Mount the snapshot database - this needs the control_files
> parameter pointing to the location of the clone file system.
> 4. Rename the redo logs - SQL script does this.
> 5. Rename the datafiles - we use RMAN to catalog the datafile
> copies and then do a switch to copy.
> 6. Backup the existing controlfile and restore the controlfile
> <--- this is the important bit as it gives you a normal control
> file and not a standby controlfile.
> 7. Recover until cancel on the database using backup controlfile.
> 8. Open database with reset logs.
>
> I have this scripted up - the database piece is independent of the
> method used to provide the copy of the datafiles.
>
> I can send through the scripts if you are interested.
>
> Regards,
>
> Mark Burgess
>
>
>
>
> On 11 February 2014 at 8:39:13 am, Don Seiler (don_at_seiler.us
> <mailto://don_at_seiler.us>) wrote:
>
>> OK I've played a little more with the duplicate steps (using
>> target-less duplication with BACKUP LOCATION specification) and
>> so far it seems to only work with backups, not with datafile
>> copies. I'd like to keep using the datafile copy method as it
>> saves time, and our database is nearly 25Tb. Creating a "backup
>> as copy" of the database and remounting the disks saves us from
>> having to backup and then restore the datafiles which would take
>> far too much time.
>>
>> Like I said, this method works great but so far requires a
>> primary controlfile. It's really only an extra step when we do
>> the refresh from the standby side but I'd like to see if we can
>> do it without having to do that step, just to simplify and script
>> as much as possible.
>>
>>
>> On Sat, Feb 8, 2014 at 1:46 PM, Don Seiler <don_at_seiler.us
>> <mailto:don_at_seiler.us>> wrote:
>>
>> Version is 11.2.0.3. I'll explore the duplicate options.
>> Anyone done it with data file copies target than backup set?
>> I'd rather not do active our require connection to primary or
>> standby other than the initial backup as copy.
>>
>
>
>
> --
> Don Seiler
> http://www.seiler.us

-- 


------------------------------------------------------------------
This email is intended solely for the use of the addressee and may
contain information that is confidential, proprietary, or both.
If you receive this email in error please immediately notify the
sender and delete the email..
------------------------------------------------------------------


--
http://www.freelists.org/webpage/oracle-l

Received on Tue Feb 11 2014 - 00:03:17 CET